Sie sind auf Seite 1von 26

Scottish Journal of Political Economy, DOI: 10.1111/sjpe.12065, Vol. 62, No.

1, February 2015
2015 Scottish Economic Society.

MAKING MONEY OUT OF FOOTBALL


Stephanie Leach* and Stefan Szymanski**

A BSTRACT

In the US, most economists argue that professional sports teams are profit-maximising businesses, but it is a widely held view in Europe that professional football clubs are not run on a profit-maximising basis. This belief has important
implications for the impact of widely-advocated policy measures, such as revenue
sharing. This paper looks at the performance of 16 English football clubs that
acquired a stock exchange listing in the mid-1990s. If the European story is true,
we should have observed a shift toward profit-maximising behaviour at these
clubs, under the assumption that investors were attracted to these football clubs
to earn a positive return. This paper finds no evidence of any shift in the behaviour of these 16 clubs after flotation. This result is consistent with the view that
football clubs in England have been much more oriented toward profit objectives
than is normally assumed.
Those clubs which have oated to become public companies
Manchester United, Newcastle United, Aston Villa, Chelsea,
Tottenham now have as their principal objective the making
of money for their shareholders.
David Conn, The Football Business, p. 154
I

I NTRODUCTION

In North America it is commonplace, especially among economists, to think


of the owners of professional sports teams as prot maximisers (Fort and
Quirk, 1995). In Europe, however, this assumption has been treated somewhat
sceptically. In an inuential paper, Sloane (1971) argued that a plausible characterisation of the owners of football clubs is as utility maximisers subject to
a budget constraint, where utility is largely associated with success on the
pitch. Reasons for this view include the perceived lack of protability of football clubs and the opinions expressed by club ocials. In some countries,
football clubs are organised as sporting associations which have no shareholders, but in England all professional clubs are limited companies, and most
have been so for around 100 years. Empirical and theoretical research exists that
*UEFA
**University of Michigan

25

26

STEPHANIE LEACH AND STEFAN SZYMANSKI

attempts to test the competing hypotheses of prot and utility maximisation in


sports, but this literature neither oers a rm conclusion nor clearly establishes evidence supporting one hypothesis over the other.
In order to add to the existing literature on objective functions in sport, this
study focuses on sixteen English football clubs that came to be traded on the
London Stock Exchange in the mid-1990s. For the most part, public trading of
these clubs arose through share placings and oers for sale of up to 100% of the
share capital. The main hypothesis is the following: If the directors of these
clubs were acting as utility maximisers prior to their otation, then otation
should have brought about a signicant change in the clubs objectives, assuming that investors in publicly quoted corporations are interested primarily in
nancial returns. To test the main hypothesis, this paper examines several
dierent aspects of performance for these sixteen clubs before and after their
otation. The results indicate that changes in the measured performance of
these clubs do not seem to be consistent with a shift toward more prot-oriented
objectives. That is, otation appears to have had little or no observable eect
on the maximising behaviour of these 16 clubs that raises capital in the stock
market. This paper explores possible interpretations of the empirical results.
II

T HE I MPACT

OF

F LOTATION

The significance of objectives for league policy


The identication of a rms objective function is central to understanding its
behaviour, and this is more than usually crucial when it comes to understanding sports clubs and leagues. Members of sports leagues typically enter into a
wide range of restrictive agreements such as revenue sharing, limitations on
players spending (salary caps and roster limits), and restrictions on player
mobility. These restraints, or so the team owners claim, are necessary to preserve competitive balance without which the leagues product will become
unattractive. Antitrust authorities have in general been persuaded by this line
of argument. However, critics such as Fort and Quirk (1995) and Vrooman
(2000) have argued that these restraints will be tend to raise prots, that this
is the true motive for their adoption by owners, and that the impact on competitive balance will be negligible or non-existent. The assumption of prot
maximisation is critical to the validity of these claims with respect to competitive balance, as has been shown in work of Kesenne (1996, 2000).
Consider, for example, the case of collectively sold broadcast rights. In the
North American sports leagues, the income derived from collective sale is typically divided equally among the teams. What eect will collectively-sold rights
have on behaviour as compared to the alternative where teams negotiate their
own broadcast rights individually and retain the income for themselves? Let us
suppose that if rights are sold individually then there are some large-market
teams that will generate substantially more income than small-market teams. If
owners are prot maximisers, there is reason to doubt whether collective selling
will improve the competitive balance of the league, since owners are under no
obligation to spend what they receive. Thus, a small market team may receive
Scottish Journal of Political Economy
2015 Scottish Economic Society

MA KI N G M O N E Y O U T O F F O O T B A L L

27

more income under collective selling, but may not choose to spend more on
creating a successful team. Under the prot maximisation hypothesis, owners
should spend up to the point where the marginal revenue of a win equals the
marginal cost, and a xed share of broadcast income will aect neither marginal revenue nor marginal cost.1 However, if the owners are utility maximisers
whose principal interest is success on the pitch, then collective selling will
improve competitive balance. By assumption teams spend what they get on the
pursuit of sporting success, and collective selling means more spending power
for the small-market teams and less spending for the large-market teams.
Ownership and motives in English football
In this paper, we are interested in the possible change in behaviour associated
with stock market otation for UK football clubs. The ownership structure of
football clubs in the UK is signicantly dierent from the model adopted in
other countries. In most of Europe, football clubs have typically been organised as not-for-prot sporting associations. Even very large clubs, such as Barcelona and Bayern Munich, have been run as clubs in a legal sense, i.e.
controlled by members who pay an annual subscription and commercially
managed by a club committee. One of the most practical consequences of this
arrangement is that non-UK football clubs have not been able to take advantage of limited liability and, therefore, their ability to borrow has been constrained. Football clubs in England and Scotland sought to evade this
restriction as early as the 19th century. No fewer than 68 of the 92 teams in
the four English professional divisions adopted limited company status prior
to the First World War, the majority before 1900.2
The conventional view is that the ownership of a limited company resides
with the shareholders and that the shareholders are motivated by prot. However, there are plausible reasons to doubt this in the case of English football
clubs. Firstly, an analysis of shareholder lists suggests that the original subscribers were largely drawn from a clubs locality and were frequently supporters of the club; hence, the prot motive may have been tempered by an
interest in sporting success. Even shareholders with purely commercial interests (such as local brewers) may have been interested in the success of the club
from the perspective of generating income for their core businesses rather than
for any direct nancial return.3
Secondly, over time most of these clubs came to be concentrated in the
hands of a small number of wealthy individuals, usually because the limited
1
Indeed, collective selling will impair competitive balance if it leads to a disproportionate
fall in the marginal revenue from winning for the small-market team (Szymanski and Kesenne, 2004).
2
For more details on the early history of English football clubs see Mason (1980), Vamplew (1988), Tischler (1981), and Inglis (1988).
3
Morrow (1999) provides a detailed analysis of the motivation of directors with dominant
shareholdings. An approach that would be complementary to ours would be to analyse the
changes in ownership that occurred before and after otation. However, this is a huge task
in itself and beyond the scope of this paper.

Scottish Journal of Political Economy


2015 Scottish Economic Society

MA KI N G M O N E Y O U T O F F O O T B A L L

47

increase their public prole and status, one example being Alan Sugar who
owned Tottenham between 1991 and 2001 (King, 1998). In such cases oatation might bring about even more recognition (or infamy) along with funds to
increase club status. However it could also be argued that non-oating clubs
are run and invested in by a class of investors that do not want to share decision-making with other potential investors who may not view playing success
as paramount. But even those clubs that have not oated, according to King
(2003), have restructured their management practices and structures in line
with public limited companies. Individual motivations behind institutional
investors and directors may be better addressed by a sociological study than
by our economic analysis. Regardless of historical ownership, once a club
oats, it is inconceivable that it will not experience some change in behaviour.
Also, one could argue the potential for selection bias in the size of clubs
that chose to oat. All the sample clubs except Preston North End and Swansea were First Division or Premiership clubs, so measuring relative protability between these larger clubs that did oat and the smaller clubs that
didnt could be a source of bias. However if we make comparisons with clubs
that did not oat and which belonged to either Division 1 or the Premiership
(a sample of 26 clubs), then pre-1997 prots for these non-oating clubs were
actually lower, averaging 1.5m. Post-1997, these clubs had comparable
losses of 3.7m on average per season.
But there is more than one interpretation of these ndings. We can identify
the main contending explanations.
(i) All football clubs were prot maximisers before otation so that entry
onto the stock market did not lead to any appreciable change in behaviour.
(ii) The clubs that floated were prot maximisers before otation so that
entry onto the stock market did not lead to any appreciable change in
behaviour relative to the average .
(iii) Accounting prots give a poor indication of economic prots, so that the
gures cannot truly indicate any change in economic performance.
(iv) Post otation accounts of PLCs include data related to group business
activities that extend beyond the football club and are therefore not comparable to the pre-otation data.
(v) These clubs did not become prot maximisers after otation because:
a Professional investors were unable to exercise control (only small
amounts of shares were oered to the market); and/or
b Professional investors were not interested only fans bought shares.
(vi) The directors mistakenly believed that the appropriate way to operate as
a prot maximiser was to invest heavily in playing talent in the anticipation of future success generating larger prots.
All of the arguments apart from (i) imply that the data need not be inconsistent with the conventional view of club objectives. The second explanation does
not seem all that plausible, given the fact these clubs had lower protability
Scottish Journal of Political Economy
2015 Scottish Economic Society

MA KI N G M O N E Y O U T O F F O O T B A L L

29

Equation (1) is a contest success function which translates total league


expenditure on talent into winning percentage for team one.
w1

t1
t1 t2

Equation (1) is the logit function commonly used in the analysis of contests
and tournaments.7
Further, suppose that revenue is concave, increasing with wins up to some
point and then decreasing, reecting the possibility that fans care about competitive balance. Assuming asymmetry in the income that teams can generate
from wins, the prot functions for each team are:
p1 r  w1 w1  ct1 ;

p2 1  w2 w2  ct2 ;

and
where r is a measure of asymmetry and c is the (constant) marginal cost of
hiring talent. Without loss of generality, assume r > 1 so that team one has
greater revenue-generating potential than team two, which would be the case
if team one was located in a larger market. Since the owners are assumed to
value both prots and success on the pitch, we can state the objective function
as a weighted average of prot and wins. Thus for team one:
X1 a1 p1 1  a1 w1 a1 r  w1 1  a1 w1  a1 ct1 ;

where a1 is a positive number less than or equal to 1. As the weight on wins


increases (1  a1), prots must eventually decrease, since wins cost money
(via increased talent expenditure) and eventually reduce income (as league
becomes excessively unbalanced).8
We dene 2 analogously.
Given these objectives the rst-order conditions for the clubs owners are:
@X1
t2
a1 r  2w1 1  a1 
 a1 c 0;
@t1
t1 t2 2

@X2
t1
a2 1  2w2 1  a2 
 a2 c 0;
@t2
t1 t2 2

and

from which we can derive the Nash equilibrium win percentage for team one
in Equation (7).
w1

a1 a2 r  1 a2
a1 a2 r  1 a1 a2

See Nti (1997) for an analysis of dierent functional forms.


Sloane (1971) assumed that a zero prot constraint applied to the clubs, but this is not
clear given that a sugar daddy might choose to fund a clubs losses in order to achieve success. Examples of this type of conduct abound, not only in football but in all major sports.
Szymanski and Kuypers (1999) discuss some examples from football).
8

Scottish Journal of Political Economy


2015 Scottish Economic Society

30

STEPHANIE LEACH AND STEFAN SZYMANSKI

From Equation (7), it straightforward to show that the win percentage of


team one will decrease when the weight on prots (a1) increases, i.e.
@w1
a2

:
@a1 a1 a2 r  1 a1 a2 2

An owner who becomes more interested in prot will invest less in playing
talent in response to the reduced optimum winning percentage. Less investment
in talent will increase prots (otherwise the owners could have both increased
prot and success by investing more in talent prior to change in objectives).
This, then, is the predicted eect of stock market otation on the team that
oats; specically, clubs that oat on the stock market are expected to see a
decrease in winning percentage and a concomitant decrease in talent investment.
We can also consider the eect of a change in one teams objective on the
performance of the other team. Consider rst the ratio of winning percentages, which in this model equals the ratio of playing talent. At equilibrium
the expression for this is:
w1 t1 a1 a2 r  1 a2

:
w2 t2
a1

From this it is apparent that a fall in investment in talent for team two will
lead to a fall in investment in talent at team two. Since the team 2 spends less
on talent and has an increased win percentage, its prots must increase, and
therefore the change in objectives of team 1 should raise the protability of its
rival clubs.
These eects follow directly from the supposed change in objectives. Indirect consequences may follow as well if the increased scrutiny imposed by the
listing requirements cause directors to be more circumspect in their policies.
First, this may involve the avoidance of excessive risks, thus creating a more
stable earnings stream. Secondly, it may imply a shift in distribution policy
toward higher and more regular dividend payments, which are sometimes considered an important indicator of company performance by market investors.
Thirdly, it may be that company eciency is improved, so that resources are
more productive and opportunities are exploited more fully. In the case of
football clubs, otation may be associated with a higher degree of commercialism, such as raising ticket prices if it is protable to do so and increasing
club-branding eorts.
III

D ATA

During the 1980s, business-minded entrepreneurs began appearing on the


managing boards of football clubs (King, 1998). In 1983, one such businessman was Irving Scholar, who masterminded the rst oatation of an English
football club: the North London club Tottenham Hotspur. The public oering
9
Win percentage for team two is also decreasing in the weight placed by its owners on
prot. Note that when there is no asymmetry in revenue generating potential (r = 1), the
optimum win percentage depends simply on the relative weights placed on prot by the two
teams (a2/(a1 + a2)) and winning percentage is still decreasing in a1.

Scottish Journal of Political Economy


2015 Scottish Economic Society

31

MA KI N G M O N E Y O U T O F F O O T B A L L

of shares in Tottenham Football Club PLC, a holding company formed to


bypass English FA restrictions against paying excessive dividends, was part of
a strategy by Scholar to improve the nancial situation at the club. In 1989,
South London-based Millwall Football Club became the second English club
to list on the stock exchange. Millwalls strategy was to create a leisure group
which would provide additional services (bars, restaurants and other leisure
facilities) in order to buttress low turnover due to poor gate receipts. The
third English football club to oat on the stock exchange was Manchester
United in 1991. Martin Edwards became chief executive and majority shareholder by inheriting his father Louis shares. The younger Edwards was twice
approached to sell his stake in the club. The rst approach was made by Robert Maxwell in 1984, with an oer of 10 million. Then in 1989, it seemed
almost certain that Michael Knighton would become the new owner, only for
the deal to fall through due to his questionable funding methods. Eventually
on 31 May 1991, Manchester United oered over 2.5 million shares to the
public via the London Stock Exchange. A major reason for Manchester Uniteds oat in 1991 was to fund the restructuring of the Stretford End.
The huge increase in broadcasting income associated with the advent of the
Premier League and the rapid appreciation of Manchester United share created conditions in the mid-1990s such that the stock market was receptive to
new issues. Between October 1995 and October 1997, a further sixteen English
clubs obtained a listing. Table 1 gives details of the otation timeline for these
16 clubs.
Table 1
Flotation Timeline
Club
Preston North End
Chelsea
Leeds United
Queens Park Rangers
Sunderland
Sheeld United
Southampton
West Bromwich Albion
Birmingham City
Charlton Athletic
Bolton Wanderers
Newcastle United
Aston Villa
Swansea City
Leicester City
Nottingham Forest

Float date

Method

% oered/placed

October 95
March 96
August 96
October 96
December 96
January 97
January 97
January 97
March 97
March 97
April 97
April 97
May 97
August 97
October 97
October 97

Placing/oer
Introduction
Takeover and placing/oer
Placing/oer
Placing/oer
Takeover and placing/oer
Reverse takeover
Placing
Placing
Placing/oer
Reverse takeover
Oer
Placing/oer
Takeover
Introduction
Oer

86
0a
60
44
26
42
100
100
30
35
100
28
16
0b
0c
11

Chelsea FC is owned by Chelsea Village PLC in which the directors and three other interests jointly held
83.5% of the equity at the companys introduction.
Swansea City FC was purchased by Silver Shield PLC, a car windscreen replacement company.
Although located in Wales, Swansea plays in English Football League and hence is treated as an English
club.
c
Leicester City FC was acquired by Soccer Investments PLC.
b

Scottish Journal of Political Economy


2015 Scottish Economic Society

32

STEPHANIE LEACH AND STEFAN SZYMANSKI

The nancial data come from the FAME database of UK-company


accounting information which provides online records for 2.8 million public
and private UK companies for the previous 10 years. Thus, in most cases we
are able to track a clubs nancial performance for about 5 years before and
5 years after otation. FAME accounts data includes prot and loss accounts,
balance sheet items, cash ow and ratios, as well as security and price information. These data are supplemented with data on club performance based
on end-of-season rankings.
Our estimation strategy is to search for any changes in the performance of
these oated companies.10 We examine four main indicators: (i) operating
prots; (ii) league ranking; (iii) relative wage expenditures; and (iv) relative
revenues.11 The rst two variables shed light directly on any possible change
in objectives associated with otation. The last two measures should be causally related to changes in the rst two variables. For instance, increased wage
expenditures likely lead to better performance in terms of league ranking.
And, increased revenues may be reected in greater protability.
Profits and dividends
There are signicant problems associated with the use of accounting prots to
measure the nancial performance of sports businesses, as is well documented
in the US literature on the subject.12
When prot and loss statements form the basis of tax assessments, rms
have a signicant incentive to understate prots. Particular government policies, for example those related to depreciation, may create tax loopholes
which enable rms to reduce prots and legally limit their tax liability. Owners may charge expenses to the company which bear little relation to any economic services rendered, thereby legally transferring taxable income away
from the company. Alternatively, companies may be able to illegally evade
tax by exaggerating expenses.
Table 2 reports operating prots for the 15 of the 16 clubs listed in Table 1.
Almost all clubs have experienced declining prots since oatation with no
10

Since Tottenham, Millwall, and Manchester United were listed during this entire period
their performance has not been considered.
11
Operating prot indicates the day-to-day operations of a football club capturing how
well a club performs in its ordinary business. Operating prot takes into account revenues
via gate receipts, marketing and merchandising, broadcasting and other commercial activities
as well as wages, salaries, cost of goods sold and various other operating costs such as amortisation of player costs. It does not include nancing costs and exceptional items (Deloitte &
Touche, 2003). Wage spending and revenues are expressed in terms of deviations from the
divisional average for two purposes. First, given the rapid escalation of ticket prices, broadcast rights values, and player salaries, a relative measure provides a consistent basis for comparison across years. Second, in the context of a sports league an absolute indicator of
nancial performance such as prots is likely to depend on the use of inputs measured in relative terms rather than absolute terms.
12
For example, see Scully (1989) and Quirk and Fort (1992). Both of these studies draw
heavily on the work of Roger Noll who dissected the prots statements of Major League
Baseball teams on behalf of the players union in the 1980s and found that reported accounting prots signicantly understated economic prots.
Scottish Journal of Political Economy
2015 Scottish Economic Society

Scottish Journal of Political Economy


2015 Scottish Economic Society

664

42

1327

1496

1879

6287

1632
76
877
69
3274
1391
9324
8640
12847
766

Bolton
Wanderers

726
488
245
803
671
1102
3780
1081
96
2780

Birmingham
City

323
2367
2899
5775
5363
9825
2408
5197
6651
9779

Aston
Villa

3836

153

1255
1080
633
495
1175
1394
6610
85
12702

Charlton
Athletic

3014

1387
3064
1107
38
1000
783
2071
6826
13508
839

Chelsea

5622

1922

3449
2908
1378
49
9763
3164
5790
2881
5685
28498

Leeds
United

Note: Post otation data based on PLC accounts. Prots for Swansea City not reported.

5
4
3
2
1
0
1
2
3
4
5
6
Average for
years before
Average for
years after

Year relative
to otation

Table 2
Operating Profits (000)

6955

355

715
2498
1038
1130
1529
1055
5786
7998
7080

Leicester
City

6662

34

671
4019
6892
3410
8105
10494
1372
19108
5126
48

Newcastle
United

7126

629

96
1161
1479
6711
2783
10130
8781
7227

Nottingham
Forest

843

201
156
291
18
715
1569
751
1476
819
272
82

Preston
North
End

6852

1634

323
3234
1680
1297
7164
5402
9524
7905
7864
3564

Queens
Park
Rangers

4849

1371

1413
5253
551
1092
950
6409
6446
5035
3553
2804

Sheeld
United

1832

21

1555
1130
955
1465
269
3401
1133
3979
2350
5099

Southampton

1609

1281

897
1499
464
2262
2863
885
1951
6046
2336
7173

Sunderland

141

1735

3342
4141
233
310
781
880
507
892
134
1834

West
Bromwich
Albion

MA KI N G M O N E Y O U T O F F O O T B A L L

33

34

STEPHANIE LEACH AND STEFAN SZYMANSKI

club reporting a positive operating prot. Newcastle reported a cumulative


loss of 25m over this period, while Nottingham Forest reported a cumulative
loss of 36m. In general a business that runs perpetually at an economic loss
will be closed by its owners if they are prot maximisers. Several of the clubs
did in fact have to undergo a signicant restructuring. The shares of Nottingham Forest, Queens Park Rangers (Loftus Road PLC), and Leicester City
have all been suspended from the market, while the latter two clubs entered
administration in 2001 and 2002 respectively. Loftus Road is no longer a
listed company, while the shares of Leicester City remain suspended at the
date of writing. Nottingham Forest had their shares delisted in 2002 following
their failure to publish their accounts and in anticipation of a restructuring
involving a cash injection of 5m from a wealthy supporter. Swansea City,
which was taken over by a listed company in 1997 was sold to it Managing
Director for 1 during the 2000/01 season. Thus, it may be that the losses
indicated reect a genuine failure to produce an economic return. On the
other hand, Bolton has reported an operating loss in each of the last nine seasons without ling for bankruptcy, while Newcastle has paid out dividends in
each of the past ve seasons (totalling 14m) despite the size of its reported
losses.
The ability to pay dividends is generally viewed as an indicator of nancial
health, although there may be many good reasons for not paying dividends. It
makes little sense for a company with protable investment opportunities to
return internally generated funds to shareholders. Of the quoted football clubs
only six paid dividends: Aston Villa, Bolton, Newcastle, Southampton, Sunderland, and West Bromwich Albion. The total payout across those years
were available was in the region of 0.9m per club per season.
In the ve or so years prior to otation, the listed clubs in total reported
losses of 17m in aggregate, an average of 0.2m per club. In the ve or so
years since otation, aggregate losses have been 284m, an average of 3.8m
per club, around 15 times larger than before otation. Surprisingly, only
West Bromwich Albion reported improved prots on average post oatation.
The operating prots of Leicester, Nottingham, Preston North End, Queens
Park Rangers, and Sheeld Untied have decreased signicantly based on
standard t-tests. Using pre-tax prot as an additional measure, only Aston
Villa, Chelsea Village, Sunderland and West Bromwich Albion reported positive prots on average and were also the only clubs where protability
improved.13
It might be argued that protability should be compared against industry
levels. There are roughly sixty-ve clubs that did not change status during the
sample for which we have accounting data.14
In the 5 years from 1992 to 1996, these clubs reported an aggregate loss of
213m, an average loss of about 0.9m per season per club. In the period
13
That is, all other clubs experienced greater losses than before otation. Chelsea Village
also appears to have had higher prots before otation, but the series is too short to make a
reasonable comparison.
14
We exclude Manchester United, Millwall, and Tottenham.

Scottish Journal of Political Economy


2015 Scottish Economic Society

MA KI N G M O N E Y O U T O F F O O T B A L L

35

1997 to 2002, these clubs reported an aggregate loss of 665m, equivalent to


around 2.27m per season per club. Thus, it appears that clubs that oated
had much larger losses after listing, and in relative terms their losses increased
after they were listed. If prot-maximising concerns had weighed signicantly
more heavily after otation, it seems hard to believe that the directors of the
listed clubs could not have done a lot more to bring their protability into
line with average of other clubs.15
The decline in protability also seems to be reected in the changing market valuations of the clubs. The market values of clubs analysed here declined
steadily and signicantly after otation, with the exception of Charlton and
Chelsea. Furthermore, this performance contrasted sharply with that of Manchester United and the market in general until the stock market started to
decline in 2000. This is consistent with a rational valuation of football club
shares based on expected protability.
League performance
Team performance can be measured in several ways. Clubs compete in a number of sporting competitions: the domestic league, the FA Cup, the League
Cup, and at the highest level the UEFA Cup and Champions League. We
choose to use domestic league ranking as our measure of relative performance
because it is the competition within which teams play most of their matches
andclub performance over time is comparable on this basis. Table 3 reports
league rank for each of the 16 clubs in each season under analysis.
The most striking feature of the data in Table 3 is that in 12 out of 16
cases, average league performance was better in the 6 years following stock
market otation than in the 5 years before. Moreover, in three of the four
cases where clubs dropped in performance also fell into severe nancial diculties and have lost their listing (Nottingham Forest, Queens Park Rangers
and Swansea City). It seems quite likely that it is the nancial crisis at these
clubs, rather than the stock market listing, that led to the deterioration in
performance. Thus, all but one of the clubs that have retained their stock
market listing since the mid-1990s have improved their league performance.
While this suggests a quite powerful tendency towards improved performance,
some caution should be exercised given the small number of observations
involved. Eleven clubs changed position signicantly pre- and post-otation.
Birmingham, Bolton, Charlton, Chelsea, Preston North End, Sunderland, and
West Bromwich Albion signicantly improved league performance, while
Nottingham, QPR, Sheeld United, and Swansea performed signicantly
worse.16
15
There is quite a lot of variability in nancial performance. Manchester United, the largest and most protable club by far is often cited as an outlier, but omitting it from the set of
clubs whose status did not change does not alter the prole of protability that much. Without Manchester United the 9296 average is a loss of 0.2m compared to a loss of 0.7m in
the 97-01 period.
16
West Bromwichs improvement is largely associated with the last two seasons (and their
subsequent promotion to the Premier League).

Scottish Journal of Political Economy


2015 Scottish Economic Society

58
48
41
44
47
35
30
27
24
25
25
25
13
46

23

17
7
2
10
18
4
5
7
6
6
8
8
16
10

6
5
4
3
2
1
0
1
2
3
4
5
6
Average for
years before
Average for
years after

Birmingham
City

Aston
Villa

Year relative
to otation

Table 3
League rank

Scottish Journal of Political Economy


2015 Scottish Economic Society

21

48
59
48
36
25
20
21
18
26
26
23
16
17
39

Bolton
Wanderers

16

36
29
34
33
37
26
35
24
18
21
9
14
12
33

Charlton
Athletic

5
11
14
11
14
11
11
6
4
3
5
6
6
11

Chelsea

4
1
17
5
5
13
11
5
4
3
4
5
15
8

Leeds
United

15

23

26
28
26
21
25
9
10
10
8
13
20
22

Leicester
City

31
42
23
3
6
2
2
13
13
11
11
4
3
18

Newcastle
United

28

8
8
22
24
3
9
20
21
20
34
31
36
26
12

Nottingham
Forest

44

63
61
63
67
75
75
69
59
59
49
45
24
28
67

Preston
North
End

42

12
11
5
9
8
19
29
41
40
30
43
52
48
11

Queens
Park
Rangers

29

13
9
14
20
30
29
25
26
28
36
30
33
23
19

Sheeld
United

12

14
16
18
18
10
17
16
12
17
15
10
11
8
16

Southampton

16

41
40
43
34
42
21
18
23
21
7
7
17
20
37

Sunderland

78

62

65
51
59
56
66
73
88
75
69
67
88
89

Swansea

28

67
53
50
43
41
31
36
30
32
41
26
22
18
48

West
Bromwich
Albion

36
STEPHANIE LEACH AND STEFAN SZYMANSKI

MA KI N G M O N E Y O U T O F F O O T B A L L

37

Wage spending
Clubs can improve their league performance by hiring or otherwise acquiring
better players. Since there is a well-functioning market for player talent, any
improvement in player quality can only be achieved through higher wage
spending.17 Wage spending is here dened in two ratios: relative to the average wage spending of all teams in the league and also relative to average
wage spending relative to the clubs divisions. Relativity is appropriate since
it not absolute spending that produces success, but outspending your rivals.
Table 4 reports wage spending relative to the league as a whole. In 50% of
cases, wage spending relative to the league average increased post otation.
As with performance, spending relative to the league also fell at those
quoted clubs that fell into nancial diculties (Nottingham Forest, Queens
Park Rangers, and Swansea, but not Leicester City). Relative league spending also fell at Birmingham City, Preston North End, Sheeld United,
Southampton, and West Bromwich Albion. However, in these cases the relative decline was quite small. Some clubs saw very large league relative
increases in spending, notably Bolton, Charlton, Chelsea, Leeds, Leicester,
Newcastle, and Sunderland. In the cases of Bolton, Charlton, Leicester,
Newcastle and Sunderland, these were also clubs that witnessed a signicant
improvement in performance.
Table 5 shows wage expenditures relative to division averages. When
spending is compared to divisional averages, increases occurred in seven of
the oated clubs post oatation.18 Notably, three of the clubs that increased
spending on average were also promoted at some point post oatation. This
is surprising since comparatively, the wages spent would be relative against
a larger divisional average. If we compare clubs that did not change divisions over the time period (Aston Villa, Chelsea, Leeds United, and Southampton) only Chelsea increased its wages relative to other clubs in the same
division.
Revenues
Clubs oated on the stock market may choose to exploit their commercial
opportunities more eectively, e.g. through merchandising and sponsorship.
This would manifest itself in the ability the extract higher revenues from a
given level of performance. Since on average relative performance improved
post-otation, one might reasonably expect that revenues relative to the league average would improve at most if not all clubs. In fact, revenues
improved at only six clubs out of the 16 relative to the league average and at
only seven relative to the divisional averages (Table 6, Table 7).
17
This is generally true unless the club possesses some distinctive capability that enables
them to extract a better level of performance from a given player than any other club. See
Szymanski and Kuypers (1999), chapter 6 for a discussion of this possibility.
18
Queens Park Rangers is one of these however it should be noted that since 1999, Limited accounts have not been led and PLC accounts contain data for London Wasps Rugby
Club.

Scottish Journal of Political Economy


2015 Scottish Economic Society

Scottish Journal of Political Economy


2015 Scottish Economic Society

0.97

0.83

2.10

2.21

Average for
years before
Average for
years after

1.17

0.74

0.54
0.66
0.85
0.91
1.32
1.39
1.25
1.02
1.03

Bolton
Wanderers

1.15

0.69

0.65
0.72
0.69
0.64
0.70
1.06
1.25
1.59

Charlton
Athletic

3.70

1.78

1.75
1.80
2.39
3.01
3.61
3.20
4.62
4.05

Chelsea

2.78

2.23

1.96
2.24
2.29
2.43
2.76
2.47
2.36
2.24
3.00
3.85

Leeds
United

1.98

1.26

1.03
1.12
1.41
1.49
1.92
1.58
2.03
2.26
2.05

Leicester
City

2.97

2.21

1.63
1.99
2.35
2.27
2.79
3.50
3.28
3.06
3.12
2.41

Newcastle
United

1.37

2.07

1.43
1.93
2.84
1.83
2.33
1.73
1.73
1.52
1.24
0.99

Nottingham
Forest

0.35

0.33
0.40
0.38
0.33
0.35
0.35
0.29
0.42
0.41
0.36

Preston
North
End

0.94

1.46

1.39
1.59
1.44
1.41
1.37
1.09
0.96
0.77

Queens
Park
Rangers

0.83

1.12

1.15
1.27
1.13
1.01
1.07
0.94
1.19
0.92
0.62
0.57

Sheeld
United

1.27

1.36

1.28
1.44
1.41
1.53
1.13
1.03
0.92
1.23
1.50
1.44

Southampton

1.68

1.06

0.97
1.05
1.15
1.08
1.22
1.23
1.26
1.28
2.50
2.16

Sunderland

0.22

0.34

0.40
0.36
0.33
0.33
0.27
0.24
0.20
0.23
0.24
0.22

Swansea

0.61

0.71

0.74
0.71
0.69
0.68
0.75
0.67
0.64
0.59
0.71
0.53

West
Bromwich
Albion

Note: Aston Villa, Birmingham City, Charlton, Chelsea Village, Leicester City, Nottingham Forest and West Bromwich Albion gures from PLC accounts. All other clubs based on Limited Company
Accounts.

0.70
0.73
0.95
1.16
1.31
1.05
0.79
0.80
0.88
0.86

1.96
1.92
2.23
2.28
2.11
2.17
1.92
2.14
2.45
2.32

5
4
3
2
1
0
1
2
3
4
5

Birmingham
City

Aston
Villa

Year relative
to otation

Table 4
Spending relative to the League average

38
STEPHANIE LEACH AND STEFAN SZYMANSKI

Scottish Journal of Political Economy


2015 Scottish Economic Society

1.438

1.167

1.027

0.868

1.116

0.819

0.692
1.024
1.085
0.703
1.000
0.413
1.633
0.591
1.923
1.285
1.265
0.517

Bolton
Wanderers

0.813

0.800

0.876
0.700
0.854
0.769
0.785
0.855
0.423
1.572
0.606
0.609

Charlton
Athletic

1.720

0.924
0.873
1.150
1.388
1.784
1.636
2.051
1.837
1.624
0.898

Chelsea

1.191

1.196

1.233
1.139
1.168
1.206
1.173
1.256
1.062
1.005
0.896
1.112
1.468
1.472

Leeds
United

0.810

1.091

1.018
1.162
1.202
0.684
1.657
0.825
0.673
0.814
0.838
0.779

Leicester
City

1.093

1.556

1.627
1.852
2.246
1.239
1.100
1.271
1.506
1.397
1.226
1.156
0.919
0.766

Newcastle
United

1.297

1.324

1.105
0.833
1.007
3.051
0.884
1.063
0.743
2.108
0.607
1.562
1.221
0.984

Nottingham
Forest

0.877

0.647
0.845
0.658
1.333
1.301
1.302
0.992
0.985
0.804
1.384
0.500
0.598
0.957

Preston
North
End

1.219

0.740

0.737
0.810
0.828
0.761
0.681
0.625
1.349
1.167
1.176
1.108
1.155
1.488

Queens
Park
Rangers19

1.008

0.815

0.590
0.667
0.662
0.593
1.188
1.189
1.161
1.453
1.408
0.780
0.707
0.693

Sheeld
United

0.518

0.699

0.704
0.743
0.747
0.742
0.740
0.514
0.442
0.393
0.493
0.555
0.550
0.598

Southampton

1.196

1.113

0.532
1.106
1.180
1.233
1.273
1.353
0.528
1.540
1.963
0.928
0.822
0.729

Sunderland

1.189

0.896

0.857
0.721
0.864
0.830
0.723
1.100
1.178
1.322
1.498
0.845
1.091

Swansea

0.804

0.998

0.778
1.589
1.428
0.738
0.806
0.829
0.820
0.774
0.902
0.896
0.645
0.856

West
Bromwich
Albion

19

Note: Aston Villa, Birmingham City, Charlton, Chelsea Village, Leicester City, Nottingham Forest and West Bromwich Albion gures from PLC accounts. All other clubs based on Limited Company
Accounts.
Note: it should be noted that since 1999, Limited accounts have not been led and PLC accounts contain data for London Wasps Rugby Club and the values contain wage information for rugby players as
well

0.903
1.522
0.826
1.023
2.896
1.457
1.299
0.957
1.230
1.111
1.062
1.474

0.781
1.139
1.000
1.177
1.105
0.963
0.932
0.818
0.855
0.907
0.884
0.876

6
5
4
3
2
1
0
1
2
3
4
5
6
Average for
years before
Average for
years after

Birmingham
City

Aston
Villa

Year relative
to otation

Table 5
Spending relative to the Division average
MA KI N G M O N E Y O U T O F F O O T B A L L

39

Scottish Journal of Political Economy


2015 Scottish Economic Society

0.81

0.78

2.25

2.79

Average for
years before
Average for
years after

1.02

1.07

0.31
0.92
0.95
2.08
0.89
1.49
1.05
0.90
0.66

Bolton
Wanderers

1.14

0.50

0.55
0.43
0.53
0.50
0.55
1.36
0.91
1.74

Charlton
Athletic

3.51

2.00

2.00
2.17
2.55
3.52
3.69
4.61
3.16

Chelsea

3.38

2.48

2.06
2.93
2.60
2.33
2.72
2.59
2.68
2.96
3.95
4.69

Leeds
United

1.91

1.33

1.53
1.10
1.17
1.53
1.34
2.02
1.82
1.99
2.02
1.80

Leicester
City

3.77

2.88

1.28
1.92
3.18
3.90
4.12
4.65
4.64
3.58
3.50
3.38

Newcastle
United

0.98

2.44

2.73
2.60
2.93
1.62
2.33
1.68
1.07
1.42
0.74
0.70

Nottingham
Forest

0.25

0.30
0.28
0.29
0.24
0.26
0.25
0.26
0.26
0.28
0.29

Preston
North
End

0.71

1.19

0.97
1.41
1.16
1.21
1.04
0.95
0.61
0.58

Queens
Park
Rangers

0.49

0.73

1.03
1.33
0.00
0.68
0.62
0.57
0.76
0.47
0.38
0.34

Sheeld
United

1.27

1.51

1.33
2.09
1.36
1.59
1.17
1.08
1.18
1.12
1.31
1.48

Southampton

1.23

1.59

1.33
2.09
1.36
1.59
1.17
1.08
1.18
1.12
1.31
1.48

Sunderland

0.16

0.32

0.32
0.30
0.28
0.19
0.21
0.14
0.16
0.17
0.16
0.15

Swansea

0.58

0.76

0.72
0.79
0.71
0.71
0.57
0.53
0.52

West
Bromwich
Albion

Note: Aston Villa, Birmingham City, Charlton, Chelsea Village, Leicester City, Nottingham Forest and West Bromwich Albion gures from PLC accounts. All other clubs based on Limited Company
Accounts.

0.51
0.69
0.70
1.10
1.06
0.89
0.79
0.71
0.79
0.82

1.80
2.24
2.44
2.05
2.73
2.57
3.02
2.92
2.79
2.42

5
4
3
2
1
0
1
2
3
4
5

Birmingham
City

Aston
Villa

Year relative
to otation

Table 6
Revenue relative to the League average

40
STEPHANIE LEACH AND STEFAN SZYMANSKI

Scottish Journal of Political Economy


2015 Scottish Economic Society

1.704

1.300

0.990

0.916

1.135

1.212

1.453
1.539
0.863
1.094
1.504
0.816
1.370
0.497
1.846
1.536
1.251
0.544

Bolton
Wanderers

0.789

0.655

0.545
0.560
0.649
0.681
0.839
0.775
0.786
0.466
1.561
0.594
0.536

Charlton
Athletic

2.199

0.818

0.774
0.854
0.827
0.906
1.036
2.871
2.622
2.683
1.964

Chelsea

1.188

1.098

1.048
1.048
1.307
1.158
0.960
1.066
0.960
0.930
1.014
1.276
1.599
1.122

Leeds
United

0.651

1.439

2.299
1.461
1.387
0.631
2.108
0.747
0.627
0.682
0.654
0.615

Leicester
City

1.270

1.789

1.615
1.922
2.552
1.421
1.609
1.615
1.721
1.604
1.223
1.131
1.152
1.239

Newcastle
United

1.098

1.401

0.796
1.385
1.158
3.484
0.670
0.914
0.623
1.530
0.487
1.260
1.337
0.874

Nottingham
Forest

0.845

1.145

0.854
1.168
0.835
1.347
1.520
1.492
0.950
0.978
0.908
1.068
0.537

Preston
North
End

0.998

0.493

0.408
0.494
0.631
0.517
0.498
0.408
1.467
0.872
1.015
0.772
1.041
1.292

Queens
Park
Rangers

0.783

0.689

0.501
0.522
0.594
0.454
1.082
0.980
0.873
1.095
0.830
0.642
0.645
0.701

Sheeld
United

0.477

0.690

0.811
0.677
0.931
0.605
0.654
0.460
0.399
0.482
0.382
0.425
0.505
0.592

Southampton

1.758

1.434

0.592
2.595
1.321
1.091
1.378
1.629
0.579
2.566
3.555
0.937
0.965
0.766

Sunderland

1.102

0.893

1.097
0.831
0.928
0.637
0.833
1.029
1.435
1.371
1.267
0.765
1.006

Swansea

1.013

1.273

0.812
1.699
1.909
0.835
1.149
1.234
1.087
1.017
0.998
0.908
0.999
1.142

West
Bromwich
Albion

Note: Aston Villa, Birmingham City, Charlton, Chelsea Village, Leicester City, Nottingham Forest and West Bromwich Albion gures from PLC accounts. All other clubs based on Limited Company
Accounts.

1.350
1.710
0.911
0.837
3.750
1.668
1.365
1.136
1.245
1.346
1.558
1.215

1.024
0.913
0.998
1.087
0.846
1.072
0.953
1.036
0.999
0.901
0.827
0.817

6
5
4
3
2
1
0
1
2
3
4
5
6
Average for
years before
Average for
years after

Birmingham
City

Aston
Villa

Year relative
to otation

Table 7
Revenue relative to the Division average
MA KI N G M O N E Y O U T O F F O O T B A L L

41

42

STEPHANIE LEACH AND STEFAN SZYMANSKI

IV

E CONOMETRIC A NALYSIS

The analysis thus far has been discussed in terms of simple averages. These
shed light on the proposition that otation shifted football club owners away
from utility maximisation toward prot maximisation given that such a
change in objectives is likely to lead to an increase in prots and a relative
decline in performance on the eld. However, another approach is to look at
the underlying causal relationships. The rst causal mechanism that underlies
the analysis in this paper is that league performance is determined by the
quality of players hired in a competitive market so that in general higher
player expenditure leads to better league performance. The second link is that
better performance will generate increased revenue as teams attract fans, sponsorship and other income as a result of increased success. This is essentially
the model proposed and estimated in Szymanski and Smith (1997). Each team
chooses a level of investment in playing talent to meet its target level of performance and prot given their underlying objectives and capabilities. We can
write:
Pit ai bwit ;

10

Rit ci dPit :

11

and
P is league rank, w is wage expenditure relative to the average and R is revenue relative to the average. The a and c parameters represent intrinsic dierences in terms of productivity (the eciency of turning player spending into
performance) and revenue generating capacity (from a given level of support).
Each team then has an objective function that is a weighted average of prots
and performance:
Xit kpit 1  kPit ;

12

so that if, for example, k = 1, the club cares only about prot. Here we ask
whether otation might change the underlying causal relationship as well as
the weighting on prot. In eect, we test to see whether a and c are aected
by otation. This might be because a stock market listing is a more eective
discipline on company managers and hence they become more productive,
either in their ability to generate playing performance from a given investment
(a) or to generate income from success (c). Note that otation, since it raises
income from the otation proceeds, should at least increase c in the short run.
Dynamic Modelling and the Error Correction Mechanism
The data available here is an unbalanced panel, which is characterised by a
relatively small time dimension (T ranges from 4 to 10, with average T = 8.8)
but a large number of clubs (N = 86). In economics there are many relationships that are dynamic in nature and a major advantage of panel data is that
we are better equipped to examine the dynamics of a relationship. Hence, the
question we are interested in is essentially a dynamic one of the adjustment,
which takes place in a club over time to otation. However, due to small T
Scottish Journal of Political Economy
2015 Scottish Economic Society

MA KI N G M O N E Y O U T O F F O O T B A L L

43

and large N, we therefore have a very well known problem in panel data estimation, which was rst outlined by Nickell (1981) that under these circumstances OLS dynamic panel data estimation is subject to considerable bias.
We therefore employ the GMM estimation technique proposed by Arellano
and Bond (1991) and Arellano and Bover (1995) to estimate dynamic panel
data models. Essentially these techniques build up a recursive varying set of
instruments which provide good small sample performance even in the face of
relatively short time periods (T), a good survey of these techniques may be
found in Baltagi (1995).
Given that actual outcomes in football will often deviate substantially from
planned results, the most natural approach to estimating these relationships is
using an error correction model. The parameters from the error correction
specication allow us to make inferences about the long-term equilibrium and
short run adjustments towards this equilibrium. The long run structure is indicated by the coecients on the level terms whereas the short term adjustments
are captures in the dierenced terms of the error correction model. We also
allow for the fact that our explanatory variables are predetermined variables;
hence we use the lagged variables as valid instruments suggested by the
Arellano and Bond study, dierenced at period t  1. Our two estimating
equations are:
DRit ci b1 Rit1 b2 DRit1 b3 Pit1 b4 DPit1 b5 Qit1
b6 DQit1 b7 Dit1 b8 DDit1 b9 PRit1 b10 DPRit1
b11 RELit1 b12 DRELit1 et

13

DPit ai d1 Pit1 d2 DPit1 d3 wit1 d4 Dwit1 d5 Qit1


d6 DQit1 d7 Dit1 d8 DDit1 d9 PRit1 d10 DPRit1
d11 RELit1 d12 DRELit1 gt
where revenues (R), wage expenditure (w) (both in orthogonal deviations)
and league performance (P) are expressed in logs, Q is a dummy variable
that indicates periods when clubs are listed on the stock market, D indicates the league division in which the team plays, PR is a dummy indicating winning promotion in the current season and REL is a dummy
indicating being relegated in the current season. The division variable will
account for the level of competition in the league on performance and also
the market size of the club with respect to revenue generation. Promotion
and relegation dummies will also account for the movement between the
divisions with respect to performance and also the generation (or loss) of
income when a team is promoted (or relegated). The Q dummy will take a
value of 1 the season after a club has oated on the stock market. For
example, Birmingham City oated in March 1997; hence Q would have a 0
for the 1996/97 season and a 1 for the subsequent seasons. Parameter estimates are reported in Table 8. The rst three columns report estimates for
the revenue equation, the last three columns reports estimates for the performance equation.
Scottish Journal of Political Economy
2015 Scottish Economic Society

DRi,t

Scottish Journal of Political Economy


2015 Scottish Economic Society

Pi,t1

wi,t1

Ri,t1

Variablesa in levels
Constant

DRELi,t1

DPRi,t1

DDi,t1

DQi,t1

DPi,t1

0.039
(0.024)

0.021
(0.048)
0.113
(0.026)

0.174
(0.021)
0.053
(0.063)
0.051
(0.054)
0.044
(0.058)
0.010
(0.058)

Variablesa in dierences
DRi,t1
0.145
(0.040)
Dwi,t1

Regression Number
Dependent variablea:

Table 8
Regression results

1.62

4.29

0.452

0.165

0.753

0.949

0.843

0.822

3.62

t-statistic

(1)

0.044
(0.024)

0.023
(0.057)
0.123
(0.027)

0.173
(0.021)
0.064
(0.066)
0.052
(0.053)
0.043
(0.054)
0.011
(0.059)

0.149
(0.041)

DRi,t

1.83

4.51

0.403

0.181

0.805

0.982

0.972

0.817

3.58

t-statistic

(2)

0.049
(0.026)

0.686
(0.099)
0.527
(0.042)

0.006
(0.020)
0.012
(0.062)
0.060
(0.046)
0.069
(0.050)
0.005
(0.053)

0.009
(0.046)

DRi,t

1.88

12.6

6.93

0.098

1.37

1.31

0.192

0.319

0.207

t-statistic

(3)

0.322
(0.047)
0.461
(0.841)

0.758
(0.175)

0.032
(0.091)
0.068
(0.054)
0.019
(0.116)
0.144
(0.097)
0.099
(0.093)
0.066
(0.094)

DPi,t

5.48

6.85

4.34

0.700

1.06

1.49

0.167

1.260

0.346

t-statistic

(4)

0.331
(0.052)
0.463
(0.085)

0.773
(0.188)

0.019
(0.090)
0.066
(0.085)
0.037
(0.123)
0.158
(0.098)
0.112
(0.096)
0.052
(0.094)

DPi,t

5.44

6.35

4.11

0.558

1.17

1.62

0.300

1.23

0.219

t-statistic

(5)

0.099
(0.119)
0.906
(0.080)

2.983
(0.303)

0.117
(0.090)
0.110
(0.042)
0.067
(0.107)
0.153
(0.080)
0.103
(0.083)
0.163
(0.079)

DPi,t

11.3

0.831

9.84

2.05

1.23

1.91

0.627

2.63

1.30

t-statistic

(6)

44
STEPHANIE LEACH AND STEFAN SZYMANSKI

Scottish Journal of Political Economy


2015 Scottish Economic Society

0.25
0.27
144.3
(0.18)
(1.01)

759

0.886

0.025
(0.029)
0.074
(0.030)
0.279
(0.079)
0.303
(0.066)

4.58

3.53

2.48

t-statistic

DRi,t

(1)

wi;t
t
w

 
ri;t
rt

 

759

0.41
0.52
477.7
(0.20)
(0.09)

0.099
(0.080)
0.345
(0.117)
0.741
(0.121)
0.997
(0.128)

DPi,t

734

7.80

6.10

2.95

1.24

t-statistic

(4)

0.41
0.52
502.7
(0.13)
(0.08)

0.081
(0.098)
0.343
(0.117)
0.725
(0.121)
0.984
(0.126)

DPi,t

734

7.83

5.98

2.94

0.826

t-statistic

(5)

0.56
0.48
738.8
(1.09)
(1.11)

0.097
(0.129)
0.822
(0.120)
1.019
(0.110)
1.104
(0.116)

DPi,t

734

9.52

9.26

6.84

0.755

t-statistic

(6)

wit is total company wage expenditure of club i in year t, rit is company turnover, and

759

4.10

5.72

4.42

1.21

t-statistic

(3)

pi;t
93pi;t

0.45
0.25
384.7
(1.52)
(1.54)

0.076
(0.063)
0.194
(0.044)
0.420
(0.073)
0.270
(0.066)

DRi,t

; Pi;t  log

4.47

3.68

2.40

1.34

t-statistic

; Ri;t log

0.26
0.27
150.4
(0.35)
(0.78)

0.046
(0.034)
0.074
(0.031)
0.282
(0.077)
0.299
(0.067)

DRi,t

(2)

pit is league rank on measured from 1 to 92, treating rst place in Division One of the Football League as rank 21, rst place in Division Two as 45, and so on. AR(1) and AR(2)
are tests of rst and second order serial correlation in the errors, distributed as standard normal. Figures in parenthesis are robust standard errors.
b
Regressions are as follows:
(1) Revenue on Performance with no time dummies or xed eects.
(2) Revenue on Performance with time dummies but without xed eects.
(3) Revenue on Performance with time dummies and xed eects.
(4) Wage on Performance with no time dummies or xed eects.
(5) Wage on Performance with time dummies but without xed eects
(6) Wage on Performance with time dummies and xed eects.
c
Observations are smaller for the Performance on Wage Equation due to some Clubs ling abbreviated accounts where they are classied as small companies and do not have to
disclose as much information such as remuneration.

Note: a Variables are dened as follows: Wi;t log

Observationsc
R2
r
WALD (joint)
AR (1)
AR (2)

RELi,t1

PRi,t1

Di,t1

Qi,t1

Regression Number
Dependent variablea:

Table 8 (Continued)

MA KI N G M O N E Y O U T O F F O O T B A L L

45

46

STEPHANIE LEACH AND STEFAN SZYMANSKI

We are interested primarily in the sign and signicance of the quoted variables. In an error correction model the terms specied in dierences specify
the way in which a given variable inuences the adjustment toward equilibrium and the levels terms dene the underlying long term equilibrium relationship. The most important result therefore is that the variable dening stock
market otation is insignicantly dierent from zero in each of the regressions
reported suggesting that stock market otation has no long-term impact on
the performance of the club. In other words, quoted teams are not expected
to generate more revenue in the long term from a given league position or to
generate a better league position from a given wage expenditure relative to the
average. However, given that the clubs that oated are mostly in the Premiership and First Division, we also ran identical error correction models adjusting for divisional averages. These results were almost identical to those
reported in Table 8 with indicating that even comparing the listed clubs with
their closest industry peers, being listed did not impact long term or short
term revenue generation. The rst of these is perhaps most surprising, since
many would have expected quoted clubs to exploit commercial opportunities
of success more eciently. One interpretation of this result is that all teams
exploit commercial opportunities fully, regardless of ownership. Another qualication we might add is that the adjustment period is longer than our panel
allows. The estimates of the dynamic terms tell a slightly dierently story. In
the wage-performance and performance-revenue equations, the dynamic terms
are insignicant, suggesting that there was not even a short-term adjustment
brought about by otation. We also tested for slope eects concerning the
quotation dummy and found them to be insignicant indicating that oatation does not have an interactive eect on performance or revenues.
V

D ISCUSSION

AND

C ONCLUSIONS

Interpreting the ndings of this paper requires some caution. On the face of it
there appears to have been a decline in protability accompanied by increase
in relative spending and league performance among clubs that oated some
shares on the market in the mid-1990s. One might question whether these
results, given the size of the sample, are statistically signicant, but the main
fact is that the expectation, based on economic theory, that prots should
increase and league performance decline following otation, does not seem to
be supported by the data.
First we must address the potential for sample selection bias. Did the clubs
that oated, already hold a disposition towards prot maximisation? According to King (2003) the massive increase in turnover during the 1990s necessitated the creation of new business structures. Due to increased trans-national
competition, commercialisation it is imperative for clubs to remain competitive in the transfer market. Directors may also have been inclined to list
shares in order to sustain a clubs nancial position and maintain value and
competitiveness (King, 2003). It can be argued that many of the shareholders
and directors of clubs (listed or not) invested in football clubs as a means to
Scottish Journal of Political Economy
2015 Scottish Economic Society

MA KI N G M O N E Y O U T O F F O O T B A L L

47

increase their public prole and status, one example being Alan Sugar who
owned Tottenham between 1991 and 2001 (King, 1998). In such cases oatation might bring about even more recognition (or infamy) along with funds to
increase club status. However it could also be argued that non-oating clubs
are run and invested in by a class of investors that do not want to share decision-making with other potential investors who may not view playing success
as paramount. But even those clubs that have not oated, according to King
(2003), have restructured their management practices and structures in line
with public limited companies. Individual motivations behind institutional
investors and directors may be better addressed by a sociological study than
by our economic analysis. Regardless of historical ownership, once a club
oats, it is inconceivable that it will not experience some change in behaviour.
Also, one could argue the potential for selection bias in the size of clubs
that chose to oat. All the sample clubs except Preston North End and Swansea were First Division or Premiership clubs, so measuring relative protability between these larger clubs that did oat and the smaller clubs that
didnt could be a source of bias. However if we make comparisons with clubs
that did not oat and which belonged to either Division 1 or the Premiership
(a sample of 26 clubs), then pre-1997 prots for these non-oating clubs were
actually lower, averaging 1.5m. Post-1997, these clubs had comparable
losses of 3.7m on average per season.
But there is more than one interpretation of these ndings. We can identify
the main contending explanations.
(i) All football clubs were prot maximisers before otation so that entry
onto the stock market did not lead to any appreciable change in behaviour.
(ii) The clubs that floated were prot maximisers before otation so that
entry onto the stock market did not lead to any appreciable change in
behaviour relative to the average .
(iii) Accounting prots give a poor indication of economic prots, so that the
gures cannot truly indicate any change in economic performance.
(iv) Post otation accounts of PLCs include data related to group business
activities that extend beyond the football club and are therefore not comparable to the pre-otation data.
(v) These clubs did not become prot maximisers after otation because:
a Professional investors were unable to exercise control (only small
amounts of shares were oered to the market); and/or
b Professional investors were not interested only fans bought shares.
(vi) The directors mistakenly believed that the appropriate way to operate as
a prot maximiser was to invest heavily in playing talent in the anticipation of future success generating larger prots.
All of the arguments apart from (i) imply that the data need not be inconsistent with the conventional view of club objectives. The second explanation does
not seem all that plausible, given the fact these clubs had lower protability
Scottish Journal of Political Economy
2015 Scottish Economic Society

48

STEPHANIE LEACH AND STEFAN SZYMANSKI

than their peers prior to otation. Moreover, this does not explain why they
would have improved their league performance by spending more on players
while presiding over declining prots. The third explanation appears weak since
whatever problems there may be in inferring the level of economic prots from
accounting prots, it is reasonable to believe that changes in accounting prots
are a good indicator in changes in economic prots over a reasonable period of
time for a large enough sample of businesses, absent signicant changes in the
accounting rules. One problem with this argument relates to the fourth explanation if the group business has more opportunities to shelter prots earned
elsewhere after otation then it may well be the case that listed companies have
an incentive to report larger losses. However, in most cases football was the primary business activity of the listed entity, and there are no cases of clubs
becoming part of much larger commercial empires as in the US.20
The fth explanation has some merits. As Table 1 shows, several clubs
oated a relatively small percentage of the stock, limiting the scope for the
market in general to put pressure on the performance of the directors
(although this story carries with it the implication that the directors were failing in their duciary duties, a serious allegation). Where small amounts of
stock were on oer, it may well have been the fans who were most likely to
buy. However, there is plenty of evidence that institutional shareholders were
signicant buyers at otation of many of these clubs, and indeed it was the
perception that this was the case that gave rise to many complaints from fans
about the commercialisation of football (see e.g. Conn (1997)). Morrow
(1999) reports that at its 1997 accounting year end 124 institutional shareholders owned almost 60% of the ordinary shares in Manchester United.
However, it may be that the institutions quickly deserted the newly oated
clubs once they realised that they were unlikely to see a reasonable return on
their investment. Few clubs ever saw their market value rise above the level
posted in the rst month of trading and most saw quite rapid declines in value
in the early months after otation. If this reects institutions selling o their
shareholdings, it is unclear why they should have given up on the idea quite
so quickly. A better picture of what happened could be constructed from an
analysis of shareholder lists.
The sixth explanation is one that also might be consistent with the market
valuation data. Directors may have gambled on improving performance with
a view to exploiting the very rapid growth in media income during the period.
The escalation of player salaries in general during this period was a reection
of this growth, and it may have appeared to be an individually rational strategy to invest relatively heavily in the late 90s with a view to obtaining a larger
share of a larger pot in the new millennium. An example of this approach
appears to be the performance of Leeds United, which invested heavily and
gambled on achieving success not only in the Premier League but also in the
UEFA Champions League. They did in fact succeed in reaching the semi-nal
20
In 1999, the UK competition authority blocked the takeover of Manchester United by
the Sky broadcasting organisation, eectively prohibiting media ownership.

Scottish Journal of Political Economy


2015 Scottish Economic Society

MA KI N G M O N E Y O U T O F F O O T B A L L

49

of the latter competition in 2001, only to fail to qualify for the following
season and found themselves unable to fund their collection of star players.
They then became forced sellers of large amounts of player talent.
It seems unlikely that any one explanation will furnish a conclusive explanation of the relative performance of the football clubs that oated in the
mid-1990s. However, the data does at least provide a serious challenge to the
received view that football clubs in England were utility maximisers rather
than prot maximisers. If a utility-maximising club oats stock on the market
the most natural implication is a shift upward in prot and downward in on
the pitch performance, almost exactly the opposite of what seems to have
occurred. Not only did prots fall and performance improve on the pitch, but
the econometric evidence suggests that the reason for this change was that the
oating clubs simply spent the otation proceeds on players. While it is not
impossible to construct alternative stories to explain the data while maintaining the conventional view that football clubs are utility maximiser, at the very
least the explanations seem somewhat strained. The alternative view that
football clubs have always been prot maximisers, in England at least,
deserves some consideration.
A CKNOWLEDGEMENTS
We thank to Dirk Nitzsche for assistance with data collection, Paul Levine
for guidance on the econometrics and Todd Jewell for reviewing the
manuscript. We thank Peter Sloane and seminar participants at the CARR
Outreach workshop on business history for helpful comments.
Refe re nc es
ARELLANO, M. and BOND, S. R. (1991). Some tests of specication for panel data: Monte
Carlo evidence and an application to employment equations. Review of Economic Studies, 58, pp. 27797.
ARELLANO, M. and BOVER, O. (1995). Another look at the instrumental variables estimation
of error component models. Journal of Econometrics, 68, pp. 2951.
BALTAGI, B. H. (1995). Econometric Analysis of Panel Data. Chichester: John Wiley and
Sons.
CHEFFINS, Brian (1998). Sports teams and the stock market: A winning match? UBC Law
Review, 32, pp. 27191.
CONN, D. (1997). The Football Business: Fair Game in the 90s? Edinburgh: Mainstream Publishing.
DAWSON, P., DOBSON, S., and GERRARD, B. (2000). Estimating coaching eciency in professional team sports: Evidence from English Association football. Scottish Journal of
Political Economy, 47, pp. 399421.
Deloitte & Touche (2003). Annual Review of Football Finance. Manchester, UK: Deloitte &
Touche.
FORREST, D. and SIMMONS, R. (2002). Team salaries and playing success in sports: A comparative perspective. Sport
okonomie, 4, pp. 22136.
FORT, R. and QUIRK, J. (1995). Cross subsidization, incentives and outcomes in professional
team sports leagues Journal of Economic Literature, XXXIII, 3, pp. 126599.

Scottish Journal of Political Economy


2015 Scottish Economic Society

50

STEPHANIE LEACH AND STEFAN SZYMANSKI

HALL, S., SZYMANSKI, S. and ANDREW Z. (2002). Testing causality between team performance
and payroll: The cases of major league baseball and English soccer Journal of Sports
Economics, 3, 2, pp. 14968.
INGLIS, S. (1988). League Football and the Men Who Made It. London: Willow Books.

KESENNE
, S. (1996). League management in professional team sports with win maximizing
clubs. European Journal for Sport Management, 2, 2, pp. 1422.

KESENNE
, S. (2000). Revenue sharing and competitive balance in professional team sports.
Journal of Sports Economics, 1, 1, pp. 5665.
KING, A. (1998). The End of the Terraces, The Transformation of English Football in the
1990s. London, UK: Leicester University Press.
KING, A. (2003). The European Ritual, Football in the New Europe. Aldershot, UK: Ashgate
Publishing Limited.
MASON, T. (1980). Association Football and English Society, 18631915. Brighton, UK: Harvester Press.
MORROW, S. (1999). The New Business of Football. London: Macmillan.
NICKELL S. J. (1981). An investigation of the determinants of manufacturing employment in
the UK. Discussion paper No. 105, Center for Labor Economics, London School of
Economics.
NTI, K. O. (1997). Comparative statics of contests and rent-seeking games. International
Economic Review, 4359.
QUIRK, J. and FORT, R. (1992). Pay Dirt: The Business of Professional Team Sports. Princeton, NJ: Princeton University Press.
SCULLY, G. (1989). The Business of Major League Baseball. Chicago: University of Chicago
Press.
SLOANE, P. (1971). The economics of professional football: The football club as a utility maximizer. Scottish Journal of Political Economy, 17, 2, pp. 12146.
SZYMANSKI, S. and KESENNE, S. (2004). Competitive balance and gate revenue sharing in team
sports. The Journal of Industrial Economics, 52, 1, pp. 16577.
SZYMANSKI, S. and KUYPERS, T. (1999). Winners and Losers: The Business Strategy of Football. London: Viking Books.
SZYMANSKI, S. and SMITH R. (1997). The English football industry, prot, performance and
industrial structure. International Review of Applied Economics, 11, 1, pp. 13553.
TISCHLER, S. (1981). Footballers and Businessmen. New York: Holmes & Meier.
VAMPLEW, W. (1988). Pay up and play the game: Professional sport in Britain 18751914.
Cambridge, UK: Cambridge University Press.
VROOMAN, J. (2000). The economics of American sports leagues. Scottish Journal of Political
Economy, 47, 4, pp. 36498.

Date of receipt of nal manuscript: 23 September 2014

Scottish Journal of Political Economy


2015 Scottish Economic Society